A digital image scaling method suitable for an embedded system. The digital image scaling method is performed by determining the respective sizes of minimum processing units of a source image and a target image so that the source image and the target image are divided into the same number of minimum processing units in the respective lateral and longitudinal direction according to a scaling ratio of the source image size to the target image size; obtaining integer weight values, each weight value corresponding to each of the points uniformly divided between two adjacent pixels, being a value of the range of 0.about.2.sup.n (n being a natural number), the sum of the weight values between two adjacent pixels being 2.sup.n; and producing the target image by obtaining target pixel values of each of the minimum processing units of the target image corresponding to the minimum processing units of the source image by using a source pixel value reflected on a corresponding target pixel value, with the weight value corresponding to the difference with the source pixel.